differential organization of the local immune response in ......human nonprogressive tuberculoma and...

Active Local Immune Response during Latent TB Infection • JID 2005:192 (1 July) • 89

M A J O R A R T I C L E

Differential Organization of the Local ImmuneResponse in Patients with Active CavitaryTuberculosis or with Nonprogressive Tuberculoma

Timo Ulrichs,1,2 George A. Kosmiadi,3 Sabine Jorg,1 Lydia Pradl,1 Marina Titukhina,4 Vladimir Mishenko,3

Nadya Gushina,3 and Stefan H. E. Kaufmann1

1Department of Immunology, Max-Planck-Institute for Infection Biology, and 2Department of Medical Microbiology and Infection Immunology,Institute for Infection Medicine, Charite University Medicine, Campus Benjamin Franklin, Berlin, Germany; Departments of 3Immunologyand 4Thoracic Surgery, Central Tuberculosis Research Institute, Moscow, Russian Federation

Background. In 90% of all cases, Mycobacterium tuberculosis infection results in latency rather than activedisease, with the pathogen being contained within granulomatous lesions at the site of primary infection. Failureof this containment leads to reactivation of postprimary tuberculosis (TB). The regional immune processes thatsustain the delicate balance with persistent M. tuberculosis, however, remain unclear.

Methods. We compared activation statuses, biological functions, and interactions of host immune cells inhuman nonprogressive tuberculoma and active cavitary tuberculous lung tissue.

Results. Dissection of early granuloma formations revealed differential cellular distribution and activationstatuses of distinct cell types in different regions relative to the central caseotic caverna or the tuberculoma intuberculous lung tissue. In patients with tuberculoma with latent infection, distant parts of lung tissue exhibitedstrong vascularization and profound proliferative activity, indicating that continuous immune defense is requiredfor mycobacterial containment, which is absent in cavitary tuberculous lung lesions.

Conclusions. We conclude that differential regulation of the local immune response is crucial for the con-tainment of M. tuberculosis and that a continuous antigen-specific cross talk between the host immune systemand M. tuberculosis is ensured during latency. This activation requires sufficient supply of nutrients and well-coordinated structural organization, both of which are lost during reactivation of TB.

Even in the 21st century, tuberculosis (TB) accounts

for an annual death toll of ∼2 million individuals. One-

third of the world’s population (i.e., 2 billion people)

are infected with Mycobacterium tuberculosis. Of TB in-

fections, 90% remain latent and do not transform into

active clinical disease [1, 2]. Latency describes the ab-

sence of clinical symptoms, sputum negativity for acid-

fast bacilli, and the nonprogression of x-ray–dense lung

lesions [3, 4]. The pathogen persists in the face of a

functioning immune system and survives within pro-

Received 17 November 2004; accepted 1 February 2005; electronically published27 May 2005.

Financial support: European Union Framework Program 6 (to S.H.E.K. and T.U.);Bundesministerium fur Bildung und Firschung, competence networks on “Patho-genomics” and “Structural Genomics of M. tuberculosis” (to S.H.E.K.).

Reprints or correspondence: Dr. Timo Ulrichs, Max-Planck-Institute for InfectionBiology, Schumannstr. 21/22, 10117 Berlin, Germany ([email protected]).

The Journal of Infectious Diseases 2005; 192:89–97� 2005 by the Infectious Diseases Society of America. All rights reserved.0022-1899/2005/19201-0014$15.00

fessional phagocytes in granulomatous lesions. In recent

years, dormancy of M. tuberculosis has received increas-

ing attention, leading to dissection of the enzymatic ap-

paratus required for adaptation to the changed environ-

mental conditions. These include activation of alternative

pathways in the citrate cycle to allow metabolism of lipids

in the necrotic center of granulomas [5, 6], adaptation

to low oxygen pressure [7], and up-regulation of genes

involved in stress responses, such as sigma factors [8, 9]

and PGRS [10]. Thus, dormancy of the pathogen has

been characterized intensively [3]. In contrast, the ques-

tion remains whether latency is caused by camouflage

due to metabolic quiescence of the pathogen or whether,

alternatively, latency is the result of a continuous cross

talk between the host immune system and persisting

pathogens that is mediated by antigen-specific T cells

and executed by mononuclear phagocytes.

The tuberculin skin test used to detect patients new-

ly infected with active TB is based on the delayed-type

90 • JID 2005:192 (1 July) • Ulrichs et al.

hypersensitivity (DTH) response, which is induced by activated

antigen-specific T cells migrating to the site of intradermal

application of antigen. A positive DTH response, as measured

by a tuberculin skin test, is supposed to represent a specific

immune response to mycobacterial antigens in the infected

human host; however, it does not distinguish between latent

infection and active TB (except for miliary TB and Landouzy

sepsis, which produce a negative DTH response) or between

M. tuberculosis infection and bacille Calmette-Guerin (BCG)

vaccination or contact with environmental mycobacteria [11].

Characterization of antigen-specific reactions of lymphocytes

from peripheral blood mononuclear cells (PBMCs) from pa-

tients with TB versus those from infected but healthy patients

or naive individuals revealed differential T cell responses to

specific cell-wall [12] and secreted antigens (e.g., early secreted

antigenic target [ESAT]–6 [13, 14]). However, none of the pe-

ripheral antigen-specific T cell responses describes unequivocal

correlates of protective immunity to TB. Thus, the relevance

of data obtained from PBMCs to the regional immune response

operative at the site of mycobacterial containment in patients

with TB can be validated only by comparative analysis of

PBMCs with lung-derived mononuclear cells.

The present analysis aims at defining latent versus active TB

at the tissue site of primary infection in humans. To define

correlates of protection within the local immune response, we

investigated lung tissue from progressive and nonprogressive

disease states (i.e., from patients with active cavitary TB un-

dergoing surgery to decrease the load of M. tuberculosis, which

is often multidrug resistant [MDR]) and compared it with tis-

sue obtained from patients undergoing surgery for unrelated

reasons, in whose lungs tuberculoma lesions were detected and

removed to prevent reactivation of active TB later in life. Im-

munohistological analyses and in vitro immunological assays

indicated a highly active state in patients with tuberculoma,

detected by Ki-67 staining and follicle-like organization, and a

distinct histopathological pattern in the granulomatous walls

of patients with cavitary TB.

MATERIALS AND METHODS

Tissue specimens. Because of extensive lung disease and prev-

alence of MDR isolates, patients with TB in Russia frequently

undergo elective thoracic surgery. Explorative thoracic surgery

is also performed to distinguish between tumors and tuber-

culous lesions and to determine the stage of tuberculous lung

destruction. Destroyed regions are removed within the ana-

tomical borders of lung segments from 2 different groups: (1)

those with progressive disease (i.e., active cavitary TB, Ziehl-

Neelsen [ZN]–positive sputum, and exhibition of all symptoms

of clinically active TB) and (2) those with nonprogressive le-

sions (i.e., tuberculoma with distinct x-ray–detectable, circular

lesions, without any sign of disease activity or access to the

bronchial system). Lung-tissue samples were obtained from 21

patients at the Department of Thoracic Surgery of the Central

Tuberculosis Research Institute: 11 patients with tuberculoma

(mean � SD age, years; 5 women and 6 men) and38.4 � 13.7

10 patients with cavitary TB (mean � SD age, 34.8 � 10.0

years; 4 women and 6 men) (table 1) [15]. Immediately after

surgery, removed lung tissue was either fixed in 4% parafor-

maldehyde overnight for histological analysis or treated with

collagenase and DNase for cell isolation (see below). All pro-

cedures involving patients were fully reviewed and approved

by the appropriate ethical review boards in Moscow and Berlin.

Informed consent was obtained from each patient in this study.

Immunohistological staining of cell-surface markers, pro-

liferating cells, and mycobacterial antigens. After deparaf-

fination, tissue sections were washed with H20, incubated in 1

mmol/L EDTA buffer at 2 bars for 5 min, and then washed

with 5% Tween 20 in PBS (TBS). Sections were incubated with

mouse monoclonal antibodies against CD20 (1:100; CM004B;

Biocarta), CD31 (1:25; CM131B; Biocarta), CD68 (1:100;

CM033B; Biocarta), or Ki-67 (1:75; M7240; DAKO) for 2 h at

room temperature. Ki-67 binds to the nuclei of proliferating

cells without any preparatory treatment of the tissue sections.

Antibody binding to surface markers and proliferation (Ki-67)

were detected by use of the same secondary antibody system.

After washing in TBS, sections were incubated for 1 h with goat

anti–mouse polyclonal antibody labeled with alkaline phospha-

tase (115-055-146; Dianova).

For detection of mycobacteria, tissue sections were blocked with

goat serum (1:100) for 10 min. Sections were then stained with

a rabbit polyclonal antiserum against M. bovis BCG (pAbBCG) (1:

1000; B0124; DAKO) [15].

Photography. All images were captured by use of a Leica

DMLB microscope fitted with a Hitachi HV-C20A 3CCD video

camera. To maintain comparability between slides, light pa-

rameters were optimized for ZN and pAbBCG staining and

then kept constant for all subsequent slides. Images were saved

by use of DISKUS (version 42034; Hilgers) and processed by

use of PowerPoint software (Microsoft).

Preparation of cells from lung tissue. Samples of surgically

removed lung tissue (∼1–5 cm3) were cut into small pieces and

incubated in complete medium (RPMI 1640, 10 mmol/L HE-

PES buffer, 200 mmol/L l-glutamine, and 5 U/mL strepto-

mycin-penicillin [all from Gibco-BRL]) containing collagenase

D (1088866; Roche) and DNase (AMP-D1; Sigma), for 3 h at

37�C with 5% CO2. Cells were isolated from tissue debris by

use of a thin mesh. After washing several times with Ca2+-free

PBS, cells were purified by Ficoll-Paque (Pharmacia) density

centrifugation. Viability was checked by trypan blue staining,

and homogeneity of the isolated leukocytes was assessed by mi-

croscopy. PBMCs from blood samples from patients with TB

were also isolated by Ficoll-Paque density centrifugation. Im-


Table 1. Clinical and macropathological characteristics of patients with tuberculoma versus patients with active tuberculosis (TB).

No.Age,years Sex

Tuberculomadiagnosis

Mycobacterial loadin surgically removed

lung tissue, cfu/g

ZN staining pAbBCG staining

Microscopy CultureTuberculoma

wallDistant

lungPericavity

tissueTuberculoma

wallDistant

lungPericavity

tissue

M1 36 F � � 6.4 � 105 � � + +M2 28 M � � 5.5 � 103 (+) � + +M4 24 M � � 5 � 107 � � + +M5 NK F � � 4.4 � 105 + + + +M7 30 F � � No growth (+) (+) + +M11 63 M � � No growth � �

M14 56 F � � ND � � + +M15 64 M � � ND � � + +M20 28 M � � 2.6 � 106 � � + �

M22 NK M � � 2.6 � 107 � � ++ +M24 NK F � � ND

Active TB diagnosis Cavity wall Cavity wall

M3 48 M +++ +++ 3.7 � 109 � � � + + +M8 26 M ++ + 7.5 � 105 � � � � + +M9 52 M +++ + 3.0 � 107 � � + � + +M10 25 M + + 4.8 � 104 � � � ++ + +M12 35 F ++ ++ ND ++ � � +++ ++ ++M19 35 M � + ND � � � + � �

M21 NK M + + ND � � � � � �

M23 30 F � + 1.5 � 106 ND ND NDM25 NK M + + ND � � � ++ + +M26 26 F + + ND � � � + + ++

NOTE. Surgically removed lung tissue was assessed from patients with active TB (sputum Ziehl-Neelsen [ZN] positive) and from those with tuberculouslesions but without clinical signs of TB. Quantification of mycobacteria was performed according to the Gaffky scale. ND, not determined; NK, not known.

mature dendritic cells (iDCs) were derived from adherent cells,

in accordance with standard procedures [12].

Interferon (IFN)–g ELISA. Irradiated (5000 R) iDCs (from

PBMCs or lung tissue) or isolated macrophages were plated in

U-bottom microtiter plate wells at 25,000 cells/well, in 0.2 mL

of complete medium containing 10% human male AB serum

(Sigma). Mycobacterial antigens (H37Rv sonicate and recom-

binant Ag85) were added at a final concentration of 20 mg/mL.

Detection was performed in accordance with standard proce-

dures [14].

Proliferation assays. Irradiated (5000 R) iDCs were plated

in U-bottom microtiter plate wells at 25,000 cells/well, in 0.2 mL

of complete medium containing 10% human male AB serum

(Sigma). Mycobacterial antigens (H37Rv sonicate) were added

at a final concentration of 20 mg/mL. Responding autologous

nonadherent lymphocytes were added to a density of 50,000 cells/

well. Proliferation was detected by use of [3H]-thymidine incor-

poration, in accordance with standard procedures [12, 14]. Stim-

ulation indices (SI) were calculated by the ratio of incorporated

[3H]-thymidine in the presence and absence of mycobacterial

antigens. SI 13 were considered to be significant.

Statistical analyses. The statistical significance of the results

was determined by use of the statistics program included in the

GraphPad Prism program (version 3.0; GraphPad). Mean pro-

liferative responses were derived from triplicate experiments, and

differences between the 2 patient groups were determined by use

of the unpaired t test with data on cytokine concentrations (nano-

grams per milliliter). The Mann-Whitney U test was used to

compare SI derived from proliferative responses (counts per min-

ute) and cellular contents in tissue (percentages).

RESULTS

Microbiological and clinical findings. Patients with cavitary

TB presented with ZN+ sputum, from which mycobacteria

could be cultured. Tissue specimens from patients with tu-

berculoma were ZN� and TB culture negative. However, both

groups had pulmonary densities of M. tuberculosis of 103–109

cfu/g of lung tissue, with higher densities in patients with cav-

itary TB. In some cases, the centers of tuberculoma lesions pre-

sented remarkably high densities of 105–107 cfu/g of lung tissue

(table 1), with similar properties in terms of culture growth,

colony morphology, and virulence as in the mouse model (data

not shown). No correlation was detected between mycobacte-

rial load in infected tissue and immune response. Within sur-

gically removed cavitary TB or tuberculoma tissue, M. tuber-


Table 2. Density of macrophages and lymphocytes in different regions of tuberculous human lungs.

No. TB disease

Distant lung tissuePericavity/tuberculoma

and cavity/tuberculoma wall tissue

Cell density, cfu/g Macrophages, % Lymphocytes, % Cell density, cfu/g Macrophages, % Lymphocytes, %

M5 Tuberculoma ND 68 22 ND 52 21M7 Tuberculoma 5.7 � 106 73 18 20 � 106 ND NDM11 Tuberculoma 2.1 � 106 57 31 1.0 � 106 56 27M20 Tuberculoma 2.6 � 106 48 25 ND ND NDM22 Tuberculoma 26 � 106 70 21 13 � 106 60 20.5M24 Tuberculoma ND 70 18 ND 44 54M19 Cavitary TB ND 65 17 ND 50 36M21 Cavitary TB 21 � 106 40 54 ND ND NDM23 Cavitary TB 1.2 � 106 40 30 1.5 � 106 42 42M25 Cavitary TB ND 35 49 ND ND NDM26 Cavitary TB ND 40 30 ND 42 42M27 Cavitary TB 30.5 � 106 44 38 25.4 � 106 41 35M28 Cavitary TB 30 � 106 39 35 20.9 � 106 31.2 47.4

NOTE After surgical removal, tissue was treated with collagenase, and cells were isolated and counted. Relative proportions of macrophages and lymphocytesare given in percentages of total cell counts in the 2 tuberculosis (TB) disease groups. ND, not determined.

culosis was detected by use of colony-forming units, immu-

nohistological staining (see below), and ZN staining. Tissue

was often found to be ZN� [16], especially in latently infected

tuberculoma tissue. In general, caseous cavities with access to

the bronchial system contained ZN+ mycobacteria in high con-

centrations, whereas mycobacteria-containing peripheral lung

tissue was ZN� in both patient groups.

Histopathological analysis. Clinical and microbiological

diagnoses of active or suspected latent M. tuberculosis infection

were confirmed by macrohistopathological findings in the sur-

gically removed tissue, and they defined the tuberculoma group

by tuberculoma formation that included intact cellular infil-

tration and that was restricted to a single site in the lungs and

the cavitary TB group by cavitary TB with 1 or more large

cavity containing a caseous necrotic mass with access to the

bronchial system. For this distinction, tissue was divided into

3 regions: (1) thick layer of the cavity wall, directly neighbor-

ing the caseotic mass in cavitary tuberculous tissue or form-

ing the wall of a productive granuloma in tuberculoma tissue;

(2) pericavity tissue surrounding these central lesions with

dense infiltrations; and (3) superficially normal tissue far from

the macroscopic lesions and harboring numerous small intact

granulomas. The distant parts of the surgically removed tissue

were rich in small granulomas containing M. tuberculosis or-

ganisms within the necrotic center or in antigen-presenting cells

(APCs) in the peripheral infiltrated tissue [15] (in brief, this

study by Ulrichs et al. describes the morphological correlates

of granuloma formation in M. tuberculosis–infected human

lung tissue and characterizes follicle-like active centers as the

major site of host-pathogen interaction). Both patient groups

were infected with an M. tuberculosis “Beijing W” genotype

family member, which was often MDR. Distribution of my-

cobacteria or intracellular mycobacterial material can be vi-

sualized by immunohistological staining using pAbBCG [16].

The pAbBCG+ cells, which contained mycobacterial material,

were identified in all 3 regions of tuberculous lung tissue. In

cavitary TB and tuberculoma tissue material, such cells were

found not only close to the necrotic centers but also in pe-

ripheral infiltrations.

Macrophage and lymphocyte content. To identify cellular

markers of mycobacterial containment, cell densities, as well

as macrophage and lymphocyte counts, were determined. Sev-

eral different regions (with a minimal volume of ∼5 cm3) in

distant lung tissue and pericavity tissue (according to the above-

mentioned definition) from each patient were cut out and

treated with collagenase (table 2). Tuberculoma tissue was sig-

nificantly richer in macrophages than was cavitary tuberculous

tissue ( , Mann-Whitney U test) and was less denseP p .0008

in lymphocyte infiltration in both central and peripheral regions

of the tuberculous lung tissue ( , Mann-Whitney UP p .0147

test). Further analysis of the lymphocyte subpopulations by flow

cytometry revealed similar percentages of CD4+ and CD8+ T

cells in both study groups (data not shown).

Differential vascularization. Staining with CD31 and mor-

phological identification of endothelial cells detected the rela-

tive distribution of blood vessels throughout the different tis-

sue regions. Distant lung tissue was well vascularized in pa-

tients with tuberculoma (figure 1 and table 3), whereas the

comparable region in patients with cavitary TB contained low

densities of CD31+ blood vessels (figure 1 and table 3). The

cavity wall, as well as the tuberculoma wall, contained numer-

ous vessels. Twenty-six different tissue samples from 14 patients


Figure 1. Vascularization of tuberculous lung tissue. CD31high-stained regions are predominantly found in the distant lung regions of patients withtuberculoma (left, bottom panel) and in the cavity walls of patients with cavitary tuberculosis (TB) (right, bottom panel). Distant lung tissue frompatients with cavitary TB was predominantly CD31low (left, top panel). In total, tissue samples from 8 patients with tuberculoma and 6 patients withcavitary TB were stained for CD31. Frequencies of CD31high and CD31low patterns are summarized in table 3.

Table 3. Frequencies of high and low densities of CD31 stainingin different regions of tuberculous tissue from patients with cav-itary tuberculosis (TB) and patients with tuberculoma.

Region

Cavitary TB(n p 6 patients)

Tuberculoma(n p 8 patients)

CD31high CD31low CD31high CD31low

Cavity/tuberculoma wall 4 0 NA NAPericavity/tuberculoma 6 0 5 4Distant lung 1 5 5 2

NOTE. In total, 26 different tissue samples from patients with cavitaryTB and patients with tuberculoma were stained and assessed for their CD31patterns. The table presents the no. of patient samples per tissue localization(compare with figure 1). NA, not applicable.

(6 patients with cavitary TB and 8 patients with tuberculoma)

were assessed.

Production of IFN-g in different parts of tuberculous lung

tissue. Production of IFN-g, the major Th1 cytokine [17], by

cells isolated from tuberculous lung tissue was evaluated upon

restimulation with mycobacterial antigens in vitro. Cells iso-

lated from the cavity wall and distant lung tissue were compared

with PBMCs from the same patients with TB. Overall, restim-

ulation with H37Rv sonicate resulted in high levels of IFN-g

responses, whereas Ag85 only elicited low levels of production

of IFN-g. Cells from distant lung tissue and PBMCs produced

comparable concentrations of IFN-g. In contrast, production

of IFN-g by cells from PBMCs and distant lung tissue from

both patients with cavitary TB and patients with tuberculoma

were higher than was production of IFN-g by PBMCs from

naive donors or cells isolated from lung tissue from patients

with cancer (data not shown) [18, 19]. For distant lung tissue,

however, production of IFN-g did not differ between the 2

patient groups. The most potent producers of IFN-g resided

in wall regions of cavitary tuberculous tissue, secreting signif-

icantly more IFN-g than cells from comparable regions in tu-

berculoma tissue ( , t test) (figure 2).P p .003

Active centers within the outer lymphocyte infiltration. To

determine the degree of immune activation in tuberculous le-

sions in the 2 groups, proliferating lymphocytes were identified

by Ki-67 staining. Discrete spots of proliferating cells were fre-

quently detected within peripheral lymphocyte infiltrations (fig-

ure 3). Serial sections of these regions revealed that both CD4+

and CD8+ T cells surrounded the active (Ki-67+) centers and

were also present within these centers, although to a lesser ex-

tent. Proliferating cells were surrounded by CD68+ APCs that


Figure 2. Mycobacteria-specific production of interferon (IFN)–g by cells from different lung regions. Pericavity tissue and macroscopically normaltissue from distant parts of the surgically removed lung tissue were treated with collagenase and DNase. Blood samples were drawn before surgery.After isolation and purification, cells were separated by adherence into nonadherent lymphocytes and adherent antigen-presenting cells. Equal cellnumbers of each fraction were stimulated with mycobacterial antigens in vitro. IFN-g in the supernatants was measured by ELISA. P values areindicated above and between the compared groups. PHA, phytohemagglutinin.

contained mycobacteria. To compare activities of the 2 groups,

Ki-67 patterns were assessed in the 3 tissue regions. Unique

spots of high activity (figure 3A) predominated in distant lung

tissue from patients with tuberculoma, whereas corresponding

regions in patients with cavitary TB were found to be negative

for Ki-67. Immune activation was detected at the rim of the

cavitary wall in a disseminated, nonfocused form and at a lower

level than that at the active centers of the periphery (figure 3B).

Proliferative activity in vitro. To verify the cellular activity

within parts of distant lung tissue from patients with tuber-

culoma, tissue-derived cells and PBMCs from the same patients

were restimulated in vitro with M. tuberculosis antigens. Figure

4 shows significantly higher proliferative responses to M. tu-

berculosis antigens by distant lung tissue–derived lymphocytes

from patients with tuberculoma than by those from patients

with cavitary TB ( , pericavity tissue from patients withP p .041

tuberculoma vs. that from patients with cavitary TB; Pp .093,

Mann-Whitney U test). As with production of IFN-g, prolif-

erative activity was highest at the rim of the cavity wall (figure

2). Proliferative activity of lymphocytes from tuberculous lung

tissue was consistently higher than that of PBMCs from the

same patients with TB. In contrast, in patients with cancer,

proliferative responses of lung tissue–derived cells were lower

than those of PBMCs (figure 4).

DISCUSSION

Ninety percent of all M. tuberculosis infections do not transform

into active disease [1, 2], and infected individuals can carry the

pathogen in lung lesions throughout their lifetimes without

signs of active TB, since containment at the site of infection is

highly effective. The functional and morphological correlates of

resistance against reactivation that are operative in these highly

productive tuberculomas, however, remain unclear. Principally,

latency could be caused by mycobacterial camouflage or, alter-

natively, by continuous cross talk between the host immune sys-

tem and the pathogen.

Older studies have shown that the pathogen can be completely

cleared from small lesions [20, 21]. However, tuberculoma lesions

used for the present study were larger, harbored culturable my-

cobacteria (table 1), and presented all signs of inactive, latent

infection. In a comparison of these lesions to those of patients

with active cavitary TB, cells harboring mycobacteria were de-

tected in the cavity wall, pericavity tissue, and, interestingly, in

peripheral, superficial, normal lymphocyte-infiltrated lung tissue


Figure 3. Distribution patterns of proliferating cells. Tissue samples from cavity/tuberculoma wall, pericavity/perituberculoma tissue, and distantlung tissue were stained with Ki-67 and assessed for density of Ki-67+ regions. Distant lung tissue from patients with tuberculoma revealed thehighest density (A), whereas distant lung tissue from patients with cavitary tuberculosis (TB) was mainly Ki-67�. The cavity walls of patients withcavitary TB presented a disseminated pattern of proliferating cells (B).

Figure 4. Proliferative responses to mycobacterial antigens of humanlung–derived cells compared with peripheral blood lymphocytes in vitro.Pericavity tissue and macroscopically normal tissue from distant parts ofthe surgically removed lung were treated with collagenase and DNase.Blood samples were drawn before surgery. After isolation and purification,cells were separated by adherence into nonadherent lymphocytes andadherent antigen-presenting cells. Equal cell numbers of each fractionwere stimulated with H37Rv sonicate in vitro (figure 2). Proliferativeresponses are shown in stimulation indices (SI, ratio of incorporated [3H]-thymidine in the presence and absence of mycobacterial antigens).

in both study groups (as has been shown earlier for the lung

apices by polymerase chain reaction [22]). This abundance of

mycobacteria suggests that all 3 tissue regions were recognized

by antigen-specific immune cells in both patient groups. This

finding already strongly argues against camouflage as a cause of

mycobacterial persistence [23] and is consistent with active cross

talk throughout the infiltrated tissue.

Tissue dissection revealed that distant lung parts were highly

vascularized in patients with tuberculoma, whereas the com-

parable region in patients with cavitary TB contained only low

densities of vascularization (figure 1). Although the cavity wall

contained numerous vessels, we assume that the positive cor-

relation between vascularization and outcome of TB infection

(tuberculoma vs. active TB disease) points to a critical role of

oxygen and nutrient supply to the affected tissue in sustaining

the productive cross talk between the host immune system and

M. tuberculosis that permits latency. Oxygen and nutrient supply

may support both the immune system and growth of the path-

ogen, as has been shown elsewhere [7, 24].

Another marked difference between lung tissue from patients

with tuberculoma and that from patients with cavitary TB was

the distribution of macrophages and lymphocytes as cellular

mediators and executors of mycobacterial containment, re-

spectively [1]. Since immunohistological findings do not reflect

the absolute numbers and distribution of leukocytes, cells were

counted after collagenase treatment. Tuberculoma tissue was

found to be enriched in macrophages and less dense in lym-

phocyte infiltration (table 2). Although antigen-specific T cell

responses are crucial for control of TB [1, 25, 26], sufficient

numbers of activated macrophages, rather than quantities of

infiltrating T cells, seem to be limiting. Macrophages have re-

cently been shown to function as carriers of mycobacteria to

the site of granulomatous lesions, thereby contributing to my-

cobacterial containment [27]. Macrophage activation by T cells

is mediated by cytokines, notably IFN-g. As a consequence, a

single T cell should be capable of activating numerous mac-

rophages in its vicinity. It has been well documented that mac-


rophages play an important executor role in protection against

M. tuberculosis [28, 29], and one of the pathogen’s evasion

strategies includes the numerical reduction of macrophages by

induction of apoptosis [30]. At the same time, mycobacteria

misuse macrophages as their habitat by suppressing their an-

tibacterial capacities [31]. The precise role that mononuclear

phagocytes play in active human TB remains to determined.

Among lymphocyte subsets, B cells appeared in higher con-

centrations in distant lung regions than in the cavity wall and

PBMCs (as analyzed by flow cytometry of cells isolated from

lung tissue [data not shown]). Their numbers in broncho-

alveolar fluids correlate with the state of infection [32]. The

presence of B cells in active centers of follicle-like structures

(see below and [15]) (figure 3) indicates contribution of B cells

to antigen presentation.

Restimulation of isolated cells from the 3 tissue regions in

vitro revealed profound production of IFN-g in the cavity wall,

indicating marked lymphocyte activation at this tissue site, as

has been demonstrated on the histological level elsewhere [33,

34]. A comparison of lung tissue from different patient groups,

however, demonstrated that, in contrast to data obtained with

PBMCs [12, 35], production of IFN-g did not correlate with

mycobacterial containment or control of TB infection in pa-

tients with cavitary TB, regardless of which mycobacterial an-

tigen preparation was used (high response to the sonicate and

lower response to the single antigen Ag85). Since cavity walls

and tuberculomas can be vascularized intensely (table 3), high

concentrations of IFN-g in these regions do not seem to ac-

count for either inhibition of vascularization or, subsequently,

the formation of a necrotic core, as has been found for inhi-

bition of tumor growth [36, 37].

Detection of active sites in situ within the tissue, by Ki-67

staining, revealed active centers in peripheral lung tissue from

patients with tuberculoma but not in patients with cavitary TB.

These centers appeared well organized: APCs in the center of

follicle-like cell aggregates were surrounded by T cells. In contrast,

the cavity walls of patients with cavitary TB presented activated

cells in a disseminated, nonfocused distribution. Such an un-

focused pattern suggests loss of organization, which facilitates

mycobacterial spread in patients with cavitary TB. Lower levels

of activation at the rim of tuberculous lesions confirm published

evidence of impaired immune responses to M. tuberculosis at the

luminal site of the cavity wall [38]. Tuberculoma walls, in con-

trast, mainly presented an organized structure comprising small

aggregates of highly proliferating cells (figure 3A). We conclude

that peripheral foci represent active centers where cross talk be-

tween the host immune system and the pathogen is orchestrated,

leading to productive protection in patients with tuberculoma.

This differential distribution of cellular activity could be con-

firmed by proliferation data from antigen restimulation of iso-

lated cells from the 3 tissue regions. Proliferation was significantly

higher for lymphocytes from distant lung parts from patients

with tuberculoma than for those from patients with cavitary TB,

whereas no such difference could be detected in the proliferation

patterns of lymphocytes from the peripheral blood (PBMCs)

derived from the 2 patient groups. We conclude that, on the one

hand, proliferative activities of lymphocytes from PBMCs do not

mirror the activation status at the site of primary infection in

human tuberculous lungs and that, on the other hand, active

sites in the periphery of affected tissue represent the morpho-

logical substrate where cross talk between the host immune sys-

tem and the pathogen is coordinated (figure 4).

Our results suggest that the dynamic balance between active

immune response and persistent M. tuberculosis is crucial for

long-term containment and prevention of disease [2]. We pro-

pose that the organization and continuous activation of the cell

aggregates in the distant lung tissue surrounding granuloma-

tous lesions in patients with tuberculoma represent the mor-

phological and functional correlates of an efficacious immune

response that ensures mycobacterial containment and prevents

dissemination.

References

1. Flynn JL, Chan J. Immunology of tuberculosis. Annu Rev Immunol2001; 19:93–129.

2. Kaufmann SH. How can immunology contribute to the control oftuberculosis? Nat Rev Immunol 2001; 1:20–30.

3. Ulrichs T, Kaufmann SH. Mycobacterial persistence and immunity.Front Biosci 2002; 7:d458–69.

4. Zhang Y. Persistent and dormant tubercle bacilli and latent tubercu-losis. Front Biosci 2004; 9:1136–56.

5. Honer zu Bentrup K, Miczak A, Swenson DL, Russell DG. Charac-terization of activity and expression of isocitrate lyase in Mycobacteriumavium and Mycobacterium tuberculosis. J Bacteriol 1999; 181:7161–7.

6. McKinney JD, Honer zu Bentrup K, Munoz-Elias EJ, et al. Persistenceof Mycobacterium tuberculosis in macrophages and mice requires theglyoxylate shunt enzyme isocitrate lyase. Nature 2000; 406:735–8.

7. Glickman MS, Cox JS, Jacobs WR. A novel mycolic acid cyclopropanesynthetase is required for coding, persistence, and virulence of My-cobacterium tuberculosis. Mol Cell 2000; 5:717–27.

8. Ando M, Yoshimatsu T, Ko C, Converse PJ, Bishai WR. Deletion ofMycobacterium tuberculosis sigma factor E results in delayed time todeath with bacterial persistence in the lungs of aerosol-infected mice.Infect Immun 2003; 71:7170–2.

9. Chen P, Ruiz RE, Li Q, Silver RF, Bishai WR. Construction and char-acterization of a Mycobacterium tuberculosis mutant lacking the alter-nate sigma factor gene, sigF. Infect Immun 2000; 68:5575–80.

10. Ramakrishnan L, Federspiel NA, Falkow S. Granuloma-specific ex-pression of mycobacterium virulence proteins from the glycine-richPE-PGRS family. Science 2000; 288:1436–9.

11. Schluger NW. The diagnosis of tuberculosis: what’s old, what’s new.Semin Respir Infect 2003; 18:241–8.

12. Ulrichs T, Moody DB, Grant E, Kaufmann SH, Porcelli SA. T-cellresponses to CD1-presented lipid antigens in humans with Mycobac-terium tuberculosis infection. Infect Immun 2003; 71:3076–87.

13. Brandt L, Oettinger T, Holm A, Andersen AB, Andersen P. Key epitopeson the ESAT-6 antigen recognized in mice during the recall of protectiveimmunity to Mycobacterium tuberculosis. J Immunol 1996; 157:3527–33.

14. Ulrichs T, Munk ME, Mollenkopf H, et al. Differential T cell responses


to Mycobacterium tuberculosis ESAT6 in tuberculosis patients and healthydonors. Eur J Immunol 1998; 28:3949–58.

15. Ulrichs T, Kosmiadi GA, Trusov V, et al. Human tuberculous granu-lomas induce peripheral lymphoid follicle-like structures to orchestratelocal host defence in the lung. J Pathol 2004; 204:217–28.

16. Seiler P, Ulrichs T, Bandermann S, et al. Cell-wall alterations as anattribute of Mycobacterium tuberculosis in latent infection. J Infect Dis2003; 188:1326–31.

17. Demissie A, Abebe M, Aseffa A, et al. Healthy individuals that controla latent infection with Mycobacterium tuberculosis express high levelsof Th1 cytokines and the IL-4 antagonist IL-4delta2. J Immunol 2004;172:6938–43.

18. Ulrichs T, Anding P, Porcelli S, Kaufmann SH, Munk ME. Increasednumbers of ESAT-6- and purified protein derivative-specific gammainterferon-producing cells in subclinical and active tuberculosis infec-tion. Infect Immun 2000; 68:6073–6.

19. Ulrichs T, Anding R, Kaufmann SH, Munk ME. Numbers of IFN-gamma-producing cells against ESAT-6 increase in tuberculosis patientsduring chemotherapy. Int J Tuberc Lung Dis 2000; 4:1181–3.

20. Ghon A. Der primare Lungenherd bei der Tuberkulose der kinder [Theprimary lung lesion in infant TB]. Berlin: Urban & Schwarzenberg,1912.

21. Opie LH. Tuberculous meningitis in childhood: pathogenesis, signsand symptoms, treatment and prognosis. S Afr J Lab Clin Med 1956;2:29–51.

22. Hernandez-Pando R, Jeyanathan M, Mengistu G, et al. Persistence ofDNA from Mycobacterium tuberculosis in superficially normal lungtissue during latent infection. Lancet 2000; 356:2133–8.

23. Dannenberg AM Jr. Immune mechanisms in the pathogenesis of pul-monary tuberculosis. Rev Infect Dis 1989; 11(Suppl 2):S369–78.

24. Collins HL, Kaufmann SH, Schaible UE. Iron chelation via deferox-amine exacerbates experimental salmonellosis via inhibition of the nic-otinamide adenine dinucleotide phosphate oxidase-dependent respi-ratory burst. J Immunol 2002; 168:3458–63.

25. Dannenberg AM Jr. Delayed-type hypersensitivity and cell-mediatedimmunity in the pathogenesis of tuberculosis. Immunol Today 1991;12:228–33.

26. Kaufmann SH. CD8+ T lymphocytes in intracellular microbial infec-tions. Immunol Today 1988; 9:168–74.

27. Cosma CL, Humbert O, Ramakrishnan L. Superinfecting mycobacte-

ria home to established tuberculous granulomas. Nat Immunol 2004;5:828–35.

28. Ehlers S, Holscher C, Scheu S, et al. The lymphotoxin beta receptoris critically involved in controlling infections with the intracellularpathogens Mycobacterium tuberculosis and Listeria monocytogenes. J Im-munol 2003; 170:5210–8.

29. Solomon JM, Leung GS, Isberg RR. Intracellular replication of My-cobacterium marinum within Dictyostelium discoideum: efficient rep-lication in the absence of host coronin. Infect Immun 2003; 71:3578–86.

30. Lopez M, Sly LM, Luu Y, Young D, Cooper H, Reiner NE. The 19-kDa Mycobacterium tuberculosis protein induces macrophage apoptosisthrough Toll-like receptor-2. J Immunol 2003; 170:2409–16.

31. Pancholi P, Mirza A, Bhardwaj N, Steinman RM. Sequestration fromimmune CD4+ T cells of mycobacteria growing in human macrophages.Science 1993; 260:984–6.

32. Corominas M, Cardona V, Gonzalez L, et al. B-lymphocytes and co-stimulatory molecules in Mycobacterium tuberculosis infection. Int JTuberc Lung Dis 2004; 8:98–105.

33. Fenhalls G, Stevens L, Bezuidenhout J, et al. Distribution of IFN-gamma, IL-4 and TNF-alpha protein and CD8 T cells producing IL-12p40 mRNA in human lung tuberculous granulomas. Immunology2002; 105:325–35.

34. Fenhalls G, Wong A, Bezuidenhout J, Van Helden P, Bardin P, LukeyPT. In situ production of gamma interferon, interleukin-4, and tumornecrosis factor alpha mRNA in human lung tuberculous granulomas.Infect Immun 2000; 68:2827–36.

35. Arend SM, van Meijgaarden KE, de Boer K, et al. Tuberculin skin testingand in vitro T cell responses to ESAT-6 and culture filtrate protein 10after infection with Mycobacterium marinum or M. kansasii. J Infect Dis2002; 186:1797–807.

36. Beatty G, Paterson Y. IFN-gamma-dependent inhibition of tumor an-giogenesis by tumor-infiltrating CD4+ T cells requires tumor respon-siveness to IFN-gamma. J Immunol 2001; 166:2276–82.

37. Tsuruoka N, Sugiyama M, Tawaragi Y, et al. Inhibition of in vitro an-giogenesis by lymphotoxin and interferon-gamma. Biochem BiophysRes Commun 1988; 155:429–35.

38. Kaplan G, Post FA, Moreira AL, et al. Mycobacterium tuberculosisgrowth at the cavity surface: a microenvironment with failed immunity.Infect Immun 2003; 71:7099–108.

differential organization of the local immune response in ......human nonprogressive tuberculoma and...

Documents